OUP user menu

A pilot study of combination anti-cytokine and anti-lymphocyte biological therapy in rheumatoid arthritis

A.W. Morgan, G. Hale, P.R.U.B. Rebello, S.J. Richards, H.-C. Gooi, H. Waldmann, P. Emery, J.D. Isaacs
DOI: http://dx.doi.org/10.1093/qjmed/hcn006 299-306 First published online: 19 February 2008


Background: Immunological tolerance in humans using anti-T-cell monoclonal antibodies (mAbs) may be hampered by a pro-inflammatory microenvironment. All clinical trials of such therapies in rheumatoid arthritis (RA), however, have selected patients with active disease at baseline. Concurrent neutralization of inflammation with a TNFα antagonist should maximize the potential of anti-T-cell mAbs to induce tolerance in RA.

Aim: To evaluate the safety of combining a TNFα antagonist and CD4 mAb in RA.

Design: An iterative pilot study focused on the safety of such combination therapy.

Methods: Eight poor prognosis, seropositive RA patients were treated with combined CD4 and TNFα blockade. Prolonged CD4 blockade was achieved with a humanized mAb, and TNFα blockade with a p55 TNF receptor fusion protein.

Results: There was a low incidence of classical first-dose reactions to the CD4 mAb, possibly reflecting concomitant TNFα blockade. An unusual anaphylactoid reaction was seen, however, and one patient developed a probable allergic reaction after several infusions. Skin rashes were common, as previously reported with CD4 mAb monotherapy. No serious infections were documented during follow-up, despite CD4+ lymphopenia in some patients. Most patients appeared to demonstrate improved RA disease control after the study. After 17–49 months after therapy, one patient was in remission, one remained off disease modifying anti-rheumatic drugs and five had stable disease, three on previously ineffective doses of methotrexate.

Conclusions: We report, for the first time in man, immunotherapy with a combination of an anti-cytokine and an anti-T-cell reagent. We witnessed an unusual first-dose reaction but there were no significant infectious complications.


Anti-cytokine therapies are effective agents for reducing inflammation and preventing joint destruction in patients with RA. Current data, however, suggest that long-term treatment will be necessary when using these drugs, with the attendant risk of opportunistic infection.1 In contrast, data from animal models suggests that the combination of anti-cytokine and anti-T-cell therapy may provide longer term benefit from brief courses of therapy. Thus, anti-CD4 therapy was synergistic with TNFα blockade,2 and anti-CD3 therapy provided particularly robust responses when combined with anti-TNFα therapy.3 An obvious concern of combining biological therapies, however, is that this approach may increase the likelihood of serious side effects, particularly infection and, potentially, malignancy. For example, infection was a significant problem when TNFα and IL-1 blockade were combined4 and TNFα blockade alone carries a significant risk.5,,6 An increased incidence of serious infections was also reported when co-stimulation and cytokine blockade were combined.7 On the other hand, the risks of cytokine blockade should be minimized if the duration of therapy could be reduced.

We have performed a pilot study of TNFα blockade and CD4 monoclonal antibody (mAb) therapy in a small cohort of patients with refractory RA. Because the dose and duration of treatment that might be needed to gain long-term disease control in man is unknown8 our study incorporated pragmatic, iterative and escalating dosing scenarios. Patients received between 1 and 12 weeks of therapy and some patients received repeated treatment. Despite the intensive nature of the treatment regimes, there were no examples of severe or atypical infections.

Patients and methods


Eight patients with active, seropositive RA were recruited (Table 1), following approval from the Local Research Ethics Committee. Seven patients were HLA-DRB1*0101, *0401 or *0404 positive and five patients were homozygous for the rheumatoid HLA-DRB1 ‘shared epitope’, which is associated with a poor prognosis.9 Each patient fulfilled the following disease activity criteria: >6/28 swollen joints and C-reactive protein >29 mg/l, and in addition either early morning stiffness >45 min or >6/28 tender joints. A 4-week washout period from all disease modifying anti-rheumatic drugs (DMARDs) was observed but concomitant stable doses of prednisolone (<10 mg) and non-steroidal anti-inflammatory drugs (NSAIDs) were allowed. No intramuscular or intra-articular steroids were permitted during the washout period or treatment phases of the study. Patients were assessed weekly during treatment, and then 2–4 weekly according to clinical need. Infections and other potential adverse events were recorded at each visit using a standard proforma. An American College of Rheumatology (ACR20) and ACR-N response was calculated at each follow-up visit.10

View this table:
Table 1

Demographic data and outcome at most recent follow-up

PatientDemographic dataLong-term outcome
Age (years)SexDisease duration (years)Previous failed DMARDsaTime (months) from last treatmentACR-N responsebCurrent therapy
A41F2S3683Me (20)
B41F8Au, Az, C, H, Me (10), My, P, S4960Me (10)
C36M1C, Me (15), S3666Me (20) + C
D51F4H, Me (10), P, Sc
E53F3Me (15), S36RemissiondMe (15) + C
F39F3H, Me (20), S1875Me (15)
G52F5.5H, Me (10), S1849nil
H52F5C, H, Me (17.5), S1743Me (15)
  • aAu: auranofin; Az: azathioprine; C: cyclosporin A; H: hydroxychloroquine; Me: methotrexate (dose in mg); My: myocrisin; P: penicillamine; S: sulphasalazine; bACR-N response: percentage of improvement in the number of swollen and tender joints with the same percentage of improvement in at least three of the following: patient's assessment of pain, patient's assessment of disease status, physicians assessment of disease status, C-reactive protein and health assessment questionnaire10; cThis patient was lost to follow-up and no long-term data is available; dFulfilled ACR remission criteria: five of the following requirements must be fulfilled for at least 2 consecutive months: duration of morning stiffness not exceeding 15 min, no fatigue, no joint pain (by history), no joint tenderness or pain on motion, no soft tissue swelling in joints or tendon sheaths, normal acute phase response.18

Therapeutic agents

The Oxford Therapeutic Antibody Centre supplied the biological agents. The CD4 mAb (hIgG1-CD4) was a humanized mAb of IgG1 isotype, capable of mediating antibody-dependent cell-mediated cytotoxicity in vitro.11 It had previously been administered in a pilot study to psoriasis and RA patients wherein brief, 5-day courses resulted in CD4 modulation, and transient depletion of CD4+ T-cells from peripheral blood.12 The bivalent TNF antagonist (TNFr-Ig) was a p55 TNF receptor extracellular domain fused to the hinge, CH2 and CH3 domains of human IgG1, and was designed, constructed and produced ‘in-house’ (this study was initiated prior to the widespread availability of anti-TNFα drugs). TNFr-Ig had not been used to treat RA previously but potently blocked TNFα in vivo.13

Study protocol

Pre-clinical data suggested that the achievement of therapeutic tolerance may require mAb-mediated masking of CD4 antigen on T-cells for a period of 6–12 weeks.8 Therefore, our aim was to slowly escalate our anti-CD4 dosing regime to one which might safely achieve this outcome, simultaneously controlling inflammation by TNFα blockade. We adopted an approach in which patients could receive more than one course of therapy. The patients were followed weekly and a disease flare was defined as a failure to maintain an ACR20 response on two successive weeks. The first five patients (A–E, Group 1, Figure 1) received initial treatment comprising a single dose of TNFr-Ig (50 or 100 mg) followed by four daily infusions of hIgG1-CD4 (125 mg/day) (standard combination therapy, Phase 1). They were followed weekly and they entered Phase II and received a second course of treatment comprising TNFr-Ig alone following a disease flare. Phase III (following a second disease flare) comprised another course of standard combination therapy. These three phases of the study provided an initial, basic comparison of TNFα monotherapy and combination therapy. A third flare was followed by a 6-week course of therapy comprising standard combination therapy during week 1 and then 5 weekly infusions of 100 mg hIgG1-CD4. A minimum CD4+ lymphocyte count of 0.4 × 109/l (lower limit of normal for the local laboratory) was required to progress through the various phases of treatment.

Figure 1.

Therapeutic strategy for Group 1 and ACR20 response duration. In Phase I, patients received 50 mg (patients marked with*) or 100 mg TNFr-Ig (solid arrow) followed by 125 mg hIgG1-CD4 daily for 4 days (broken arrows). Each patient is represented by a different symbol. In Phase II, patients received TNFr-Ig monotherapy, at the same dose as in Phase I. Three patients progressed to Phase III and received 100 mg TNFr-Ig followed by 125 mg hIgG1-CD4 daily for 4 days. In Phase IV, patients received 100 mg TNFr-Ig followed by 125 mg hIgG1-CD4 daily for 4 days, and then weekly infusions of 100 mg hIgG1-CD4. Patients were required to flare before progressing to a subsequent phase of the study, and to maintain a minimum CD4+ lymphocyte count of 0.4 × 109/l.

Three additional patients (F–H, Group 2) entered Phase V, which was designed to maintain CD4 blockade for 12 weeks, while continuously suppressing inflammation with TNFr-Ig. Therapy comprised TNFr-Ig (100 mg) at weeks 0, 4 and 8 and hIgG1-CD4 (200 mg weekly) from weeks 0 to 11.

Lymphocyte subset analysis

Lymphocyte subtyping (CD3+, CD3+ CD4+, CD3+ CD8+, CD19+ B cells and CD3− CD16+ or CD56+ CD16+ NK cells) was performed using flow cytometry. The antibodies used to label CD4+ cells recognized non-overlapping CD4 epitopes to hIgG1-CD4, and included QS4120 (American Tissue Type Culture Collection) for Phases I–IV and Leu-3a for Phase V.

Laboratory monitoring

Full blood count was measured weekly during therapy, urea and electrolytes and liver function tests every 2–4 weeks. Anti-nuclear antibody was measured monthly and rheumatoid factor, anti-double stranded DNA antibodies and anti-cardiolipin antibodies bi-monthly.

Anti-globulin responses

Anti-globulins to the biological agents were measured using a double capture ELISA as previously described.14 In short, ELISA plates were coated with either hIgG1-CD4 or TNFr-Ig and blocked according to standard procedures. Following incubation with test serum, plates were washed and anti-globulins detected with biotin-labelled hIgG1-CD4 or TNFr-Ig, respectively.

Skin prick testing

Test solutions were placed onto the skin and pricked into the dermis using a lancet. A 5 mm wheal from histamine (1 mg/ml, NHS pharmacy supplied) served as a positive control and no reaction to normal saline, a negative control. Antibody solutions used were hIgG1-CD4, TNFr-Ig, combination of the two and a human IgG1 anti-CD18 mAb as an isotype control. Initial tests used mAb solution at 10 μg/ml in normal saline, then 100 μg/ml if no reaction and finally 1 mg/ml. The test was terminated (positive reaction) if a wheal of 3 mm or greater than negative control was elicited after 10 min.


A comparison of adverse reactions in this study with a previous study utilizing hIgG1-CD4 was performed using Fishers exact test.



Five patients with active RA were recruited for Phases I–IV (Table 1). Four were females, mean age was 44 years and all were seropositive. Disease duration ranged from 1 to 8 years (median 3), and previous DMARD usage from 1 to 8 drugs (median 3). Three patients completed all four treatment phases. One withdrew between Phases II and III for reasons unrelated to her RA and a second patient was excluded from Phase III due to CD4 lymphopenia.

Three additional seropositive, female patients entered Phase V (Table 1). Mean age was 48 years, disease duration 3–5.5 years (median 5) and previous DMARDS used 3 or 4 (median 3). Two completed the 12-week course of treatment, but one discontinued at week 9 secondary to escalating infusion reactions (see below).



No infections were documented during the treatment period of the study, and no serious infections during the follow-up period, which lasted from 17 to 49 months (serious infection implied a requirement for parenteral antibiotics or admission to hospital).

First-dose reactions

Cytokine release syndromes or ‘first-dose reactions’ (Table 2) have been associated with the initiation of therapy with a number of therapeutic mAbs, particularly those that bind to cells.12,15,,16 In their most severe form, these can be life-threatening.17 The classical syndrome occurs 90–120 min into the infusion and comprises chills, fever, headache and, in severe cases, bronchospasm, chest tightness and hypotension. No reactions occurred with the first dose of p55-TNFr-Ig but a mild or moderate severity reaction was seen in 2/8 patients treated with hIgG1-CD4 for the first time. All patients in Phase I, however, experienced an unusual, very early reaction to the first dose of hIgG1-CD4. The reaction occurred within the first 5 min of the infusion and comprised flushing, vomiting, chest tightness, palpitations and dyspnoea, with transient hypotension additionally occurring in three patients. The symptoms reversed rapidly on stopping the infusion, and did not recur when restarted or with subsequent doses. Patients were re-exposed to hIgG1-CD4 during Phases III and IV (Figure 1). These exposures were preceded by the administration of 10 mg parenteral chlorpheniramine and no reactions were seen on these occasions. The same pre-medication was administered to patients in Phase V (who had not previously been exposed to hIgG1-CD4), but two of three patients developed early reactions. All of the CD4 mAb solutions had undergone batch testing for endotoxin and this was deemed an unlikely cause of these reactions. In a previous study using hIgG1-CD4 at a similar dose, 10/11 RA and psoriasis patients experienced a classical first-dose reaction and a single patient an atypical reaction (p = 0.004 and 0.001 for a comparison between these studies).12

View this table:
Table 2

Adverse effects recorded during and subsequent to therapy

TrialPatientEarly reactionaClassical reactionbEscalating reactions (Phase V)cRashesd
Phase IA+++
Phase VeF+
  • aEarly reaction: − none, + flushing and/or nausea/vomiting, ++ flushing and chest tightness, +++ flushing and hypotension and/or chest tightness. bClassical reaction: − none, + temperature ↑ 0.5°C and/or chills, ++ temperature ↑ 0.6–1.5°C and/or rigor, +++ temperature ↑ >1.6°C and/or hypotension. cEscalating reactions: − none, + flushing and/or nausea/vomiting, ++ flushing and chest tightness, +++ flushing and hypotension and/or chest tightness or pyrexia >38.5°C. dRashes: − none, +localized and settled spontaneously within 1 week, ++ generalized lasting several weeks. eAll patients in Phase V received 10 mg parenteral chlorpheniramine prior to the first dose of hIgG1-CD4.

Allergic reaction

A recurrent infusion reaction was seen in two patients in Phase V (patients F and G, Table 2). Transient flushing was seen with each hIgG1-CD4 infusion in both patients. However, patient G developed transient chest tightness on week 8, which was additionally associated with hypotension, rigor and fever on week 9 and, as a result of these reactions, no further infusions were given. A mild eosinophilia (0.52 × 109/l) also developed in this patient, which persisted for 8 weeks, although the eosinophil count remained above baseline for 72 weeks (Figure 2). Patient H developed mild facial oedema following the final infusion of hIgG1-CD4, which persisted for several days. This patient also developed a transient eosinophilia (Figure 2). Because these appeared to be allergic reactions, skin prick tests were performed on the three patients from Phase V and four of the patients from the earlier phases. All mounted a satisfactory reaction to histamine but no reaction to either biological agent either alone or in combination, at concentrations up to 1 mg/ml. This significantly reduced the likelihood of, but did not exclude, the development of IgE anti-globulins.

Figure 2.

Eosinophil counts of patients participating in Phase V.


Maculopapular rashes were seen in five patients during therapy, four from Phases I–IV and one from Phase V (Table 2). In three cases, the rashes were generalized and pruritic, lasting for several weeks before clearing. Skin biopsies obtained from four of these patients demonstrated mild immune cell infiltration, but with no evidence of vasculitis or immune complex deposition.

Lymphocyte depletion

In Phase I, there was a rapid fall in peripheral blood CD4+ lymphocyte count (Figure 3). At week 10, CD4 counts remained lower than baseline in four patients, but were generally within the normal range, consistent with previous experience using this mAb.12,,16 Patient B had a prolonged fall in CD4 count following Phase I, and was excluded from Phases III and IV. No further sustained reduction in the CD4 count occurred in patients re-exposed to hIgG1-CD4 in Phase III, but there was a further fall in two of three patients exposed to 6 weeks of therapy in Phase IV. Prolonged dosing with hIgG1-CD4 in Phase V also resulted in CD4+ lymphopenia, with slow recovery. There were no consistent changes in the CD8, NK cell or B cell populations during any phase of the study.

Figure 3.

CD4+ lymphocyte counts from Phases I–V.

Laboratory values

No abnormalities of renal or hepatic function were noted during the course of the study.

Anti-globulin responses

After extensive testing, we could not demonstrate an immune response against either hIgG1-CD4 or TNFr-Ig, even in the patients with escalating infusion reactions.


There was no significant change in rheumatoid factor titre and no patients developed new dsDNA or ANAs during our studies. Patients A and E had transient increases in IgM, but not IgG, anti-cardiolipin antibodies following the unopposed TNF antagonism in Phase II. These returned to baseline within 6 months and no further rises were seen during subsequent treatment courses.

Clinical outcome

This was not a formal efficacy study, and the number of patients treated was small, but the design allowed limited observations to be made. TNFr-Ig provided marked relief of RA symptoms in all patients and on each occasion it was administered, with initial improvement in the ACR-N response of 40–93% after 1 week (data not shown). Pooling all patients, irrespective of TNFr-Ig dose, median (range) ACR 20 response duration was 3 (2–10) weeks for Phase I, 1 (1–3) week for Phase II, 5 (1–6) weeks for Phase III and 6 (2–9) weeks for Phase IV (Figure 1). The shortest response durations corresponded to treatment with TNFr-Ig alone, suggestive of an additive effect or synergy between the two biological agents.

Pharmacokinetic modelling suggested that 100 mg hIgG1-CD4 was probably insufficient to maintain coating of CD4+ lymphocytes between weekly doses (Dr N. Rapson, personal communication). This resulted in the protocol adopted for Phase V, during which all three patients achieved an ACR70 response on at least one occasion. One patient has subsequently required no further treatment for her RA, except a single dose of intramuscular steroid (120 mg methylprednisolone) for a minor flare at 56 weeks. The remaining two patients flared shortly after the end of treatment at weeks 14 and 17, but subsequently responded to methotrexate at previously ineffective doses (Table 1). At 18 months, all three patients had ACR-N responses of 43–75% (Table 1).

At 3 years, of the four patients available for follow-up from Phases I to IV, one is in remission18 and three have stable disease with ACR-N responses of between 60% and 83%. One of these is taking methotrexate (MTX) at a previously ineffective dose (Table 1).


This is the first report of the use of combination anti-cytokine and anti-T-cell biological therapy in man. Our study was prematurely terminated secondary to CD4+ T-cell depletion and our conclusions are therefore based on a small patient cohort albeit with substantial follow-up. Consistent with our previous studies of lymphocytotoxic mAb therapy,19 we saw no infectious complications in the current study. In addition, there have been no serious infections during 18 months (Phases I–IV) to 3 years (Phase V) of routine clinical follow-up.

First-dose reactions were documented in the majority of patients treated but these differed qualitatively and appeared earlier than classical first-dose reactions.12,20–22 The latter are the consequence of lymphocyte-bound mAb being cross-linked, via Fcγ Receptors (FcγR), leading to cytokine secretion from the lymphocytes themselves, or from the FcγR-bearing cells.23,,24 TNF plays a crucial role in symptomatology, along with IL-6 and interferon-gamma (IFNγ).25 The prior administration of TNFr-Ig may, therefore, explain the low incidence of classical reactions compared to previous experience using hIgG1-CD4.12 More puzzling was the regular occurrence of a very early reaction, which was anaphylactoid in nature. Although reactions were not seen in Phase III, when the same patients received chlorpheniramine pre-medication, they did occur in Phase V (with a new patient cohort) despite similar prophylaxis. Reduced reactions in Phase III may therefore have reflected tachyphylaxis. Similar reactions were not seen with TNFr-Ig alone and rarely with hIgG1-CD4 alone in a previous study.12 It is possible that TNFr-Ig sensitized T-cells, with rapid mediator release following subsequent binding of hIgG1-CD4. TNF blockade facilitates signalling within RA T-cells,26 lending some credence to this hypothesis.

Two patients in Phase V became sensitized to hIgG1-CD4 infusions. In one this led to an anaphylactoid reaction, in the other to facial oedema, but both developed peripheral blood eosinophilia. We could not detect anti-globulins against either agent despite repeated testing and it remains possible that the patients became sensitized to a minor component of the mAb preparations. In animals, CD4 mAbs administered at sufficiently high doses induce tolerance to themselves.27 Furthermore, although TNFα antagonists can be immunogenic,28,,29 this outcome should be reduced by concomitant CD4 therapy.30

Lymphopenia developed following extended dosing with hIgG1-CD4 (Figure 3) whereas sustained CD4+ T-cell depletion did not follow a short course of therapy with this mAb.12

Three of our patients developed generalized and protracted rashes. Histology was non-specific, but did not suggest vasculitis or immune complex deposition, similar to reports with several other CD4 mAbs.16,31,,32

This was not an efficacy study, but all patients available for long-term follow-up (7 of 8) had substantially improved symptoms and signs compared to baseline. One was in disease remission, and one no longer required DMARD therapy. Of the remaining five patients, three were stable on previously ineffective doses of methotrexate. A formal controlled trial would obviously be required to substantiate these observations, which hint at some type of disease modulation or ‘reprogramming’.8 Immune deviation contributes to immunomodulation in animal models of immunotherapy.33,,34 The eosinophilia noted in patients G and H probably represented hypersensitivity reactions to a component of therapy, but sustained combination therapy may have driven a therapeutically beneficial Th1 to Th2 cytokine switch.

Several CD4 mAbs, with distinct properties and profiles, have been tested in early phase RA clinical trials.35 There was a disappointing lack of observed efficacy compared to pre-clinical data from animal models, but these studies were performed in the absence of therapeutic tolerance biomarkers. RA outcomes rely heavily on markers of inflammation and agents that induce therapeutic tolerance may not primarily be anti-inflammatory. Furthermore, anti-inflammatory and immunosuppressive drugs may even disrupt tolerance mechanisms. In the absence of validated tolerance biomarkers, critically important parameters such as drug dose and therapy duration will continue to be chosen pragmatically, as in our study. The ‘wrong’ choice may lead to potentially tolerogenic therapies being inappropriately and prematurely abandoned.36,,37 In contrast, used with a suitable anti-inflammatory regime and with appropriate biomarker monitoring, such therapies could yet prove to be potent immune modulators in humans.

In conclusion we report, for the first time, combination therapy in man with anti-cytokine and anti-T-cell ‘biological’ therapies. We witnessed an unusual first-dose reaction and a high incidence of skin rashes, but no infectious complications despite a degree of CD4+ lymphopenia. Efficacy outcomes were encouraging but formal, controlled trials of such protocols additionally require the identification and development of appropriate biomarkers of immune modulation.37,,38


We thank all consultants who referred patients for this trial, Sister Claire Kellett who assisted with the infusions, and the staff of the Therapeutic Antibody Centre. This work was supported by the Arthritis Research Campaign, UK. (Grant reference I0522).


View Abstract